We’ve covered the general requirements of battery packs for drones in the past, but we know that agricultural UAVs may need to cover many hectares between each charge so can afford to be slightly larger and heavier than, for example, surveillance drones which tend to stay closer to ground control. They don’t require stealth like military UAVs might, so noise is less of an issue.
Key factors in battery cells, enclosures and connectors for agricultural drones will therefore be tolerance to dust and moisture, reliability and long life. Where flights are likely to be long in duration and distance, a heavier battery with a higher power-to-weight ratio is the best selection. After all, nobody wants to trek through the Australian outback to recover a drained drone from farm land extending over a million hectares. 1
ROVs and agricultural robots (agribots) are making an increasing impression on the agricultural sector and many labour-intensive jobs are now being filled by robotic farmhands. From fruit-picking arms to manure-clearing vacuums all the way to autonomous tractors, projects can be tiny or enormous so computing requirements can differ vastly and designers need to consider these at a very early stage of agritech design.
Influencing factors include the type and amount of data being collected, speed at which that data needs to be processed and how smart the robot or product is expected to be. For example, relaying information about crops that need watering doesn’t need to be in real-time as a short delay isn’t critical, but designers may need to include AI-ready hardware that can accommodate advanced software to better understand the condition of crops or the movement of animals. And of course, this all needs to be ruggedly housed to withstand extremes in weather and conditions.
There have been some really interesting stories in the media about agribots using advanced imaging techniques to acquire more agricultural data. These include the use of hyperspectral imaging to determine crops which may need more or less fertiliser, machine learning software for image classification to identify and destroy weeds and even 3D facial recognition imaging to recognise happy pigs!
These applications all require high-resolution image capture and processing from a compact and lightweight camera, and autofocus-zoom (AFZ) cameras offer the most flexibility and control. It’s also useful to consider what sensor will be most appropriate – is a drone or robot expected to operate in low light or even at night? Is it capturing images of moving or static subjects? Is a wide field of view required, or a more narrow, precise one?
When designing for the agricultural sector, it’s important to understand how the device, vehicle or robot will be expected to share information with a control centre in order to create a product that will truly add value in the field.
Connecting tools to farmers can be done using wireless networks and 5G is promising lower latency, higher capacity and increased bandwidth. It’s designed to deliver data at up to 20 Gbps, which knocks 4G, at around 15-30 Mbps, right out of the water. It’s also able to manage a greater number of devices so networks can be larger than ever before. While 5G is still very much in the roll-out phase globally, 4G still offers decent options for less time-critical communication. Alternatively, a Mobile Ad-hoc NETwork (MANET), such as Wave Relay® from Persistent Systems, can be created to allow fully scalable and mobile communication.
To send useful info, a device needs to know where it is, and a Global Navigation Satellite System (GNSS), such as GPS, GLONASS or Galileo, is required. These rely on highly accurate Positioning, Navigational and Timing (PNT) data and several solutions have been built around PNT which include data from weather, traffic, air quality and other localised factors.
We were recently involved in submitting a bid for a UK start-up which has successfully deployed other agribot prototypes. The requirements for this model included a rugged embedded pc capable of running AI-driven software, suitable battery options, high-resolution imaging and advanced PNT capabilities.
The control unit stipulated high power requirements while the smaller, mobile unit required power over ethernet based on a Linux platform. Involving multiple business units, Steatite was able to offer rugged solutions to every challenge which have been proven in field operations and known to be highly flexible and reliable.
Our proposal included:
• Axiomtek and ADLINK (Nvidia Jetson NX based) computers, bringing high-end intelligence and offering available stock and ongoing UK support.
• Tiny Jetson industrial carrier board for powerful processing on a compact board.
• Cincoze GM-1000 – a rugged, compact and high-performance GPU computer with embedded GPU expansion.
• Positioning, Navigation and Timing (PNT) from Orolia which have the advantage of being modular so new features or functionality can be simply added via software or hardware upgrades.• Customisable lithium-ion battery and Battery Management System (BMS) to ensure optimal power deployment and long life.
• Compact and lightweight autofocus-zoom camera options from our imaging division, Active Silicon, with multiple outputs including HD-SDI, USB3, HDMI, HD-VLC™, Ethernet IP H.264 and MIPI CSI-2.
Designing and building world-class products for use in agriculture requires experience and knowledge gained over years and across multiple sectors, as the size and function of this tech can vary so greatly.
At Steatite, we’re proud to be able to offer a wealth of insight and practical advice to ensure that the diversifying applications within this industry can be significantly advanced.
Contact us to find out more about our customised, application-specific or general off-the-shelf solutions.
